Multi-Class Classification of Gastrointestinal Diseases using Deep Learning Techniques

Objective:

Currently, gastrointestinal problems claim the lives of about two million individuals every year throughout the world. Video endoscopy is one of the most important tools for detecting gastrointestinal diseases (GD). However, it has the disadvantage of producing a huge quantity of images that should be analysed by medical professionals, which demands more time. This makes manual diagnosis challenging, encouraging research into computer-assisted methods for accurately identifying all of the produced images in a short amount of time. The goal of this research is to develop anArtificial intelligence technique to classify the different diseases in gastrointenstine.

Materials and Methods:

In developing a framework for classifying gastrointestinal diseases, the proposed methodology is unique. VGG-19 and ResNet-50, two deep learning networks, were used to evaluate their ability to recognise a dataset of lower GD. To validate the proposed models, a Kvasir dataset of 3500 images was employed, comprising 500 images from each of the seven classes.

Results:

ResNet-50 had the greatest validation accuracy of 96.81%, recall of 95.28%, precision of 95%, and 94.85% of F1 score, whereas VGG-19 had the highest validation accuracy of 94.21%, recall of 94%, precision of 94.28%, and 94% of F1 score.

Conclusions:

The Convolutional neural network (CNN) was used to extract shape, colour, and texture features. Softmax assists in the classification of seven types of GD by passing these features into fully connected layers. Two models performed equally well and yielded promising results.

A state-of-art review on antenna designs for ingestible application

Research interest in ingestible Wireless Capsule Endoscopy (WCE) studies in humans showed better results than conventional invasive probe endoscopy methods. Because of the structure and the position of the small intestine, proper scanning cannot be done in the area using traditional endoscopic methods. For patient comfort, continuous developments have been suggested in capsule endoscopy designs in terms of the quality of images transmitted, capsule orientation, positioning of the capsule from outside the body, link budget analysis, impedance matching and capsule miniaturization. To improve the image quality, transmission efficiency of the antenna has to be improved. This has led to the development of many antenna structures in an ingestible capsule system. Literatures have identified Med Radio and ISM (Industrial, Scientific and Medical) band as operation bands for the WCE systems. This review aims to highlight: (1) design considerations for various antenna types, (2) miniaturization techniques, (3) operating bands, specifications and various design challenges and (4) research gap, advanced design technologies and targets of ingestible antenna system. The main aim of this paper is to tutorial the up-to-date information on the recent antenna designing techniques and challenges for ingestible system.

A stable miniaturised AMC loaded flexible monopole antenna for ingestible applications

This paper proposes an ultra-miniaturised, flexible 5.8 GHz antenna to give better performance for an ingestible wireless capsule endoscopy system. The use of an artificial magnetic conductor (AMC) unit cell in this design helps in enhancing the gain, improving the efficiency and reducing the size at resonance. This novel AMC-based layer is modelled as a radiator, and is excited by a complementary strip feed. The proposed design is printed on a flexible polyimide material of size 4.6 × 7.6 × 0.15 mm, with dielectric constant 3.5 and loss tangent 0.008. The antenna has an omni-directional radiation pattern with gain 1.64 dBi. The size is reduced drastically by bending the structure to fit inside a standard-sized ingestible capsule of 26 × 11 mm. The performance metrics of the antenna after placing inside the wireless ingestible capsule are presented with the help of an analysis of the reflection coefficient curves, radiation pattern, specific absorption rate (SAR) and field distributions. In addition, different positions and orientations of the antenna and different bending conditions are used, and these cases are solved using our AMC-based unit cell radiator. We carry out simulations using High Frequency Structure Simulator (HFSS) and measurements are made by immersing the fabricated antenna in minced meat. The results of this novel structure are compared with those of other ingestible antennas proposed in the literature. The proposed design has a SAR value of 0.82505 W/kg at around 5.8 GHz over 1 g of tissue, thereby redirecting the radiation away from the human body. This conformal AMC structure achieved better miniaturisation while maintaining the antenna resonance and SAR within permissible limits.